abq-pete

I run 3 pre clears on them while still in the original casings. If they pass, I rip them out and install them into severs. If they fail, they get exchanged. Regards, Peter

http://www.amazon.com/dp/B00829THLE STCA4000100 external unit that most likely has the ST4000DM000 5900rpm drive in it... Free shipping. Regards, Peter Now up to $149.99...

My Costco (Redwood City, CA) has the 4TB on sale again for $139.99. Regards, Peter

Yep. I bought one to have a peek inside. Mine contained a ST3000DM001. Precleared fine and works without issue in one fo my arrays.

Costcos in my area (Redwood City and Foster City, California) already have them listed for this new price...

Thanks!

I'm not sure how Supermicro is doing it with this X9SCM board but the 1st x4 slot is running off of the processor just like the two x8 slots. According to Intel, the layout should either be one x16 or two x8 only, not two x8 plus one x4 as there are only 16 lanes available. The X8SIA board gets around this using a PCI-E switch. You need to verify that the AOC-SAS2LP-MV8 will function correctly in an x4 slot. There is no documentation on the Supermicro website for this card so I cannot tell. Assuming that it does indeed work then an x4 Gen 2 PCI-E has plenty of bandwidth for 8 green drives. Also be mindful of the unraid version you use. v5b11 is the fastest release of unraid. My parity checks run in excess of 110MB/s with all green drives (21 drives total) on the X8SIA board with 3 LSI controllers. Of course as the checking approaches the inner tracks of the drives, the speed will slow down. I am testing rc3 and it seems fast but not quite as fast as b11 for me so far (prelim). Regards, Peter

I agree with you on the bandwidth numbers. Unfortunately, many people are using the AOC-SASLP-MV8 that, while an x4 card, is only PCI-E Gen 1 so it only has access to half the available bandwidth of a PCI-E Gen 2 slot. The AOC-SAS2LP, an x8 Gen 2 card, has 4 times the bandwidth of the older card. Regards, Peter

There was no fire... but also not boot. Yep, you need ECC ram as regular ram will not boot. So it is: i3 = ECC unbuffered Xeon = ECC buffered or unbuffered As a bonus, it seems that you get ECC functionality even with an i3 (at least according to reports and the boot screen identifies ECC ram). Regards, Peter

I'll test it later today...

It does NOT have an electrical x16 slot. It is only a physical x16 slot and will only support x8 speeds. You would use the "x16" and the next two x8 slots. The last x8 slot is actually an x4 slot fed off of the PCH 3420 chip (along with everything else including Lan, video, ipmi, usb, etc., etc.). You can see all this if you download the user manual. It contains a very useful diagram showing the layout. Regards, Peter

I bought my X8SIA-F board from Newegg as well though I waited for a refurb unit. When they become available, the pricing is around $145. The unit I received seemed new to me. If you can wait, it might be worth it. The processor is fine. If you decide that you want ECC ram support, you'll need a Xeon. I am toying with that now. I think the ram is fine. I use Kingston ECC ram as I might make the switch to a Xeon. http://www.newegg.com/Product/Product.aspx?Item=N82E16820139262 The SAS cables are too costly at Newegg. You can get them cheaper directly from Norco: http://www.ipcdirect.net/servlet/Detail?no=216 or buy similar cables even cheaper from my favorite cable place Monoprice: http://www.monoprice.com/products/product.asp?c_id=102&cp_id=10254&cs_id=1025410&p_id=8189&seq=1&format=2 And don't be ashamed by OCD. I like the neat layout and bandwidth capabilities of having all drives running of identical controllers on x8 slots. Although I must admit I am looking at getting one of the following: http://www.provantage.com/highpoint-technologies-rocketraid2760a~7HPTI03W.htm Good luck! Regards, Peter

Forgot to mention that the Supermicro boards support mass storage controllers to be used in all the slots and have bios options to disable the roms on the cards as well. Regards, Peter

If you are not tied to socket 1155, I've used (or am using) 3x IBM BR10i as well as M1015 cards (all x8) on the following motherboards with success: MSI 790FX-GD70 (socket AM3). This is an AMD based board so you can run low power Semprons and other more powerful AM3 socket processors. While this is an older board, the newer 890 and 990FX chipsets should be fine as well. The x90FX chipsets have plenty of PCI-e bandwidth (lots of lanes). I don't use this board other than for testing as AMD's idle state consumption is higher than Intel's. Supermicro X8SIA-F (socket 1156). This is an Intel based board. It uses a PCI switch that allows three x8 slots to be used from the x16 normally available from the processor. This is not an issue since the 16 lanes at PCI-e 2.0 have enough bandwidth to support about 32 hard drives (not SSD as they can easily saturate the bus). Beginning parity checks with 20 data drives easily exceeds 100MB/s. Socket 1155 only allows 16 lanes from the CPU, and DMI to the south bridge is already loaded so finding a board with 3 x8 close is going to be difficult. Maybe Supermicro will make an X9SIA-F. Otherwise the older X58 chipset (LGA 1366) or the new X79 (LGA 2011) offer enough PCI-e lanes. Regards, Peter

http://www.newegg.com/Product/Product.aspx?Item=N82E16816101358R Well that went fast!

Yes. I borrowed one from a friend a while back. It worked fine. I cannot recall which beta I was testing at the time though… Regards, Peter

Deal back on at 1pm California time as a Shell Shocker Deal from Newegg... Regards, Peter

Actually all 4 of your PCI-X slots share the same x8 connection from the 3210 MCH. No "rewiring" can help that. You can get a PCI-E card (AOC-SASLP-MV8 or AOC-SAS2LP-MV8) and put that in the x4 or x8 slot (respectively) and move some drives to improve your throughput. Regards, Peter

For me, the more unusual aspect of this pending announcement is that Hitachi is releasing anything new since its acquisition by Western Digital. I hope that this trend continues as I prefer the Hitachi drives to my old favorite Western Digital. I've no recent experience with Samsung but have tended to shy away since their last problems. Regards, Peter

http://www.storagereview.com/gtechnology_hitachi_shows_new_4tb_hitachi_hard_drive Looks like Hitachi is still innovating... Regards, Peter **edit** Looks like the site violated their NDA. However, my RSS feed shows the following text: ""In their latest G-RAID storage array, G-Technology by Hitachi showcased the new 4TB Hitachi 7,200RPM hard drive in a device aimed at professional A/V editing markets. The new four-platter drive is making use of a new 1TB platter design, which increases available capacity 33% compared to current-generation 3TB hard drives.""

Should'nt be the 8x bus interface an advantage, if you read from a lot of drives? (eg. rebuild the parity) PCIe V1 at 4x is 1000 MB/s, that still exceeds (or is about equal too) 8x 5400rpm drives in standard use, including parity check. hence my previous statement. you might get minor improvement. but minimal. My experience is a little different. The PCIe bandwidth numbers are theoretical. After overhead, the real-world numbers are closer to 600-700MB/s for a V1 x4. Both Hitachi and WD 2TB Green drives can push upwards of almost 130MB/s on their outer stripes. So 5 of those drives can saturate the card (the old SASLP). When you add drive 6, you'll notice that the sync speeds begin to drop. By the eighth drive, you parity speeds will max out around 60+MB/s. Moving to cards that support more lanes (x8) or PCIe V2 will reduce the bottlenecks. There are other factors such connection points (DMI, Northbridge, etc.) that also have an impact. An article that I found useful is this: http://blog.zorinaq.com/?e=10 Regards, Peter

My early Labor Day start gave me time to do some more beta testing on unRAID. Here are some parity checks speeds for the various betas, kernels and drivers. Tests were performed on the following system: MSI 790FX-GD70 3 x BR10i Controllers (all in x8 slots) 20 x 2TB drives (mix of WD and Hitachi; all 5400rpm) unRAID Linux Kernel mptsas Speed MB/s (after about 5 minutes) version version 5b8d 2.6.37.6 3.04.17 100+ 5b9 2.6.37.6 3.04.17 100+ 5b10 2.6.39.3 3.04.18 45+ 5b11 2.6.37.6 3.04.17 100+ 5b12 3.0.3 3.04.19 70+ A quote from the 5b11 release notes: ""Also, this release uses the last really stable version (for unRaid) of the linux kernel: 2.6.37.6. Something changed in the kernel starting with 2.6.38/39 which causes a throttling of I/O to hard drives as soon as you get beyond 6 hard drives accessing in parallel. Eventually I'll figure this out."" Looks like 3.0.3 is better than 2.6.39.3 but still noticeably below that of 2.6.37.6. In reviewing this thread, those that reported sync speeds close to that of 2.6.37.6 have tested only 6-9 drives in their array. Anyone else with large arrays seeing the same thing? Regards, Peter

Yes that PAL error is a b*tch. Out of four motherboards I use to test UNRAID, only one works for upgrading firmware MSI 790FX-GD70 (fail) Intel 975XBX2 (fail) Supermicro C2SEE (fail) MSI H55M-ED55 (success) Good luck

Thanks! Regards, Peter

PM Sent. Regards, Peter